1. Field of the Invention
The present invention relates in general to a method to detect the movement of an image sensor. In particular, the present invention relates to a method for detecting movement of an image sensor which, by comparing the images captured by the image sensor at different locations, finding the matched images and performing some verification processes, the movement of the image sensor is determined.
2. Description of the Related Art
Normally, the prior arts use block matching to determine the movement of an image sensor. Ideally, using this method can correctly find a matched frame block generated from the matching function, mean squared error (MSE) or mean absolute difference (MAD), then use the matched frame blocks to calculate the movement of the image sensor.
FIG. 1A and FIG. 1B show the conventional method to determine the movement of an image sensor by block matching. The frame 10A is the image captured by the image sensor in a first location, and the frame 10B is the image captured by the image sensor when the image sensor moves to a second location. Here, the size of the frame captured by the image sensor is 6×6 units, and the frame 10A is divided into 9 parts by 2×2 units. Then, the frame 10B is searched by block matching to find a frame block matching the frame block 12A in frame 10A.
However, there are actually many sources of noise, not only from image and process variation, but also from the environment of the image sensor such as temperature and illumination variation. Therefore, the matching result can only obtains a frame block 12B having a gray scale close to the frame block 12A, wherein the value of the matching function is the minimum. Then, a frame block 14B matching the frame block 14A is obtained from the image 10B by the same way.
In addition, the prior art has to compare all the frame blocks in the original images to determine the movement of the image sensor according to the frame 10A and 10B, that is, the conventional method has to fully search to get the result. Therefore, the conventional method requires using a large amount of data, so the determining rate is slow.